Factor 3: Spending Power
4.4.6 Objective 6: Market Segments found within the Youth Market
88
(convenience), 14 (anonymous), 19 (discussion and conversation), 20 (empowered), 22 (favourably), 23 (value for money) and 24 (variety). Cluster 2 also has mean values of 2 (agree) for variables 1 (responsive), 2 (reliable), 3 (salespeople), 4 (no difficulty), 5 (valuable information), 6 (satisfied), 11 (like), 15 (admire), 18 (chance) and 21 (cheapest). Cluster 2 has only two mean values of 3 (neutral) for variables 16 (exceed my budget) and 17 (compulsive shopper).
Hence, cluster 1 comprises a market segment of consumers who value social media as a promotional tool because it provides them with pleasure, convenience, anonymity, affordability, value, reliable information, sufficient information, ease in accessing this information, discussion and conversation around products and brands, a variety and wider selection of products and empowerment and encouragement to engage in purchasing behaviour through social media platforms.
Cluster 2 has predominantly mean values of 3 (neutral) and mean values ranging up to 5 (strongly disagree). This market segment of consumers are sceptical and unconvinced about the potential of social media as a promotional tool.
Cluster 1 is made up of 123 respondents, thus cluster 1 comprises the majority of the respondents in this study. Cluster 1 indicates the importance and immense value respondents place upon social media as a promotional tool.
The results from the cluster analysis also indicate that there is significant potential for social media as a promotional tool. This is supported by Stelzner (2009), and Kichatov and Mihajlovski (2010). In addition, according to research conducted by Forrester Research, social media marketing will beat e-mail, search marketing, display advertising and mobile marketing, thus demonstrating the potential of social media as a promotional tool (Tozian, 2009 cited in Kichatov & Mihajlovski, 2010).
89
factor analysis. The method used for the initial extraction of factors was the Principal Components Analysis.
Table 4.22: KMO and Barlett’s Test (n=145)
Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .895 Bartlett's Test of
Sphericity
Approx. Chi-Square 5384.226
Df 666
Sig. .000
According to the results presented in Table 4.23, a total variance of 77.646 percent is explained, which is an acceptable level (Malhotra, 1993). Eight components were extracted using the SPSS software which is illustrated by the Scree Plot in Figure 4.2. “The Scree Plot reflects Eigenvalues against the number of factors in order for extraction, thus the shape of the resulting curve is used to evaluate the cut-off point” (Hair, Anderson, Tatham & Black 1998). According to Figure 4.2, the graph elbows at 8 components, where the eigenvalue is one.
It should be noted that for any higher number of components, the eigenvalue falls below one, indicating that the variance achieved is fairly low (Malhotra, 2001). The first component illustrated in the Scree Plot indicates the dominant value which explains the maximum amount of variance which is 46 percent. The first component accounts for over half (46 percent) of the total variance of 77.646 percent. The other seven components account for much less of the total variance in comparison to the first component. In addition, from the eighth factor onwards the line is almost flat, indicating that each successive factor accounts for smaller and smaller amounts of the total variance.
90 Table 4.23: Total Variance Explained (n=145)
Component
Initial Eigenvalues Extraction Sums of Squared Loadings Total % of Variance Cumulative % Total % of Variance Cumulative %
1 16.872 45.599 45.599 16.872 45.599 45.599
2 2.835 7.663 53.262 2.835 7.663 53.262
3 2.269 6.133 59.395 2.269 6.133 59.395
4 2.096 5.665 65.060 2.096 5.665 65.060
5 1.271 3.435 68.495 1.271 3.435 68.495
6 1.189 3.213 71.708 1.189 3.213 71.708
7 1.168 3.158 74.866 1.168 3.158 74.866
8 1.029 2.780 77.646 1.029 2.780 77.646
Extraction Method: Principal Component Analysis.
Figure 4.2: Scree Plot
The data was also analysed using Varimax and Promax rotation since these two methods provide more reliable information for interpretation. In addition, by using the aforementioned methods it becomes evident how different a rotated solution is from another. The outcome of the factor rotation is reflected in Table B.1 (Appendix B).
Table B.2 (Appendix B ) which is the out come of factor anal ysis using pr omax r otation illustrates how different the rotated solution can be f rom one ot her and what is meant by a simple structure. With an oblique rotation such as promax rotation, the factors are allowed to be co rrelated w ith one anot her, whilst w ith an or thogonal r otation, such as the v arimax rotation ( shown abov e), t he factors ar e not al lowed t o be co rrelated ( UCLA R esearchers, 2010). Oblique rotations form both factor patterns and factor structure matrices (Table B.1) whereas with or thogonal r otations, the factor structure and t he factor pattern m atrices ar e
91
the same. The factor structure matrix represents the correlations between the variables and the factors and is often called the factor loading matrix (UCLA Researchers, 2010). The factor pattern matrix also represents the linear combination of the variables.
In addition, Table B.3 (Appendix B) shows that rotation has been carried out using an oblique rotation. If an orthogonal rotation was conducted (such as the varimax rotation shown in Table B.1), this table would not appear in the SPSS output. This is because the correlations between the factors are set to 0, hence with an orthogonal rotation it is assumed that the factors are not correlated. In this case, however, the factors are highly correlated allowing for correlations between the factors which is indicative of a oblique rotation (UCLA Researchers, 2010)
Eight underlying factors were extracted, which is evident in Tables B.1 and 4.24. These factors were labelled as follows: Commercial Enthusiasts, Network Commercial Information, Network Risk Takers, Network Risk Avoiders, Network Promotional Tools, Network Information Influencers, Passing Trend and Social Media’s Future.
Factor one (Commercial Enthusiasts) which addressed a respondent’s attitudes and perceptions towards social media platforms accounted for 46 percent of the 78 percent variance explained. The various items comprising this factor produced values of .880, .860, .794 and .772. This is an indication of a fairly high degree of correlation which indicates that the majority of the respondents in this study take on the role of commercial enthusiasts on social media platforms.
Factor two (Network Commercial Information) accounts for 8 percent of the 78 percent variance explained and included items with values of .790, .769 and .742. This included items on respondent’s satisfaction with the amount of information, importance and value of information and the ease of accessing this information on social media platforms. Currently, several consumers in the youth market invest much time on research prior to making purchases as they value making informed decisions. Social media platforms are a medium that the youth are increasingly turning to and depending on for being informed (Mabry, 2008).
Factor three (Network Risk Takers) accounted for only 6 percent of the total variance explained. This is an indication that a very small percentage of participating respondents in this study have assumed the role of network risk takers.
Factor four (Network Risk Avoiders) explained 6 percent of the total variance explained and included items with negative values such as -.795,-.752,-.731 and -.728 with variables such
92
as complexity, unreliability and risk. The first item, ‘purchasing through social media platforms is complex’ suggests that if purchasing through social media platforms were simple (a negative attribute rating) this would have resulted in a positive score for this item. The second item ‘payment facilities on social media platforms are unreliable’ indicates that if payment facilities on social media platforms were reliable (a negative attribute rating) this would have produced a positive score for this item. The third item, ‘I will wait until purchasing on social media platforms become safe and then purchase, rather than take chances’
indicates that if purchasing on social media platforms were safe (a negative attribute rating) this would have resulted in a positive score for this item. The fourth item ‘I do not engage in purchasing activity over social media platforms because of the risks involved’ suggests that if there was no risks (a negative attribute rating) involved in purchasing over social media platforms then this would produce a positive score for this item.
Factors five (Network Promotional Tools), six (Network Information Influencers), seven (Passing Trend) and eight (Social Media’s Future) explained less than 5 percent of the total variance explained and is therefore less significant than factors one, two, three, four and five. Factor eight (Social Media’s Future) explained only 3 percent of the 78 percent total variance explained which provided the least explanation and hence is the least meaningful in terms of the objectives of this study.
The market segments extracted from the factor analysis addresses objective 7 of the study which was to determine market segments found within the youth market. The market segments identified in the study bear a certain degree of similarity to the six social techno graphic profiles identified by Li and Bernoff (2008).
93
Table 4.24: Interpretation of Extracted Factors (n=145)
Variable Attribute Varimax